Analyzing Meaning in Big Data: Performing a Map Analysis Using Grammatical Parsing and Topic Modeling

Abstract

Social scientists have recently started discussing the utilization of text-mining tools as being fruitful for scaling inductively grounded close reading. We aim to progress in this direction and provide a contemporary contribution to the literature. By focusing on map analysis, we demonstrate the potential of text-mining tools for text analysis that approaches inductive but still formal in-depth analysis. We propose that a combination of text-mining tools addressing different layers of meaning facilitates a closer analysis of the dynamics of manifest and latent meanings than is currently acknowledged. To illustrate our approach, we combine grammatical parsing and topic modeling to operationalize communication structures within sentences and the semantic surroundings of these communication structures. We use a reliable and downloadable software application to analyze the dynamic interlacement of two layers of meaning over time. We do so by analyzing 15,371 newspaper articles on corporate responsibility published in the United States from 1950 to 2013.

Disambiguating and Specifying Social Actors in Big Data: Using Wikipedia as a Data Source for Demographic Information

“…In the dawning age of big data in text analysis (Evans and Aceves 2016; George et al 2016), social scientists, to scale up the analysis of social action, are able to automatically detect social actors and classify them as persons and organizations by using techniques such as named entity recognition (Evans and Aceves 2016; Goldenstein and Poschmann 2019; Mohr et al 2013; van Atteveldt, Kleinnijenhuis, and Ruigrok 2008). However, a continuing challenge of contemporary studies concerned with the investigation of social action on a large scale is to disambiguate social actors and to gather additional demographic information on these actors (Franzosi 1990; Goldenstein and Poschmann 2019; Sudhahar et al 2013). Specifically, social scientists continue to struggle to automatically disambiguate who is hidden behind a person’s name (e.g., a politician, an economic leader, or an athlete) or the label for an organization (e.g., a nongovernmental organization, a company, or an educational institution).…”

Conceptualising fairness: three pillars for medical algorithms and health equity

“…117 Domain experts, such as clinicians, social scientists or patient advocacy groups, have enhanced understandings of context situated bias, 114 116 118 support the curation of salient axes of difference, 119 and improve topic modelling and natural language processing models by aiding social bias detection. [120][121][122] For example, 'computational ethnography' is an approach to fairness-aware ML that emphasises the importance of a holistic understanding of any given dataset. 123 124 In sum, provenance requires more than a bias assessment that measures predictive accuracy across protected groups.…”

The extended computational case method: a framework for research design

“…It would not be an overstatement to say that topic models are one of the most successful methodological innovations associated to the rise of computational social science, given their ability to classify and code large collections of text relatively seamlessly-they offer, in a way, the forms of scalability that are desired, and promised, by computational research designs. Furthermore, the fact that topic models can now be implemented with relatively low costs and barriers of entry (via, for example, builtfor purpose software packages and libraries such as Gensim implemented in Python) means that they have diffused rapidly, being now applied to study constructs varying from meaning (Goldenstein and Poschmann 2019) and the dynamics of social movements (Lindstedt 2019) to organizational frames (Pardo-Guerra 2020) and research paradigms (Lockhart 2020).…”